distribution, formation, and seasonal variability of...

Distribution, formation, and seasonal variability of

Okhotsk Sea Mode Water

Sergey Gladyshev,1 Lynne Talley,2 Gennady Kantakov,3 Gennady Khen,4

and Masaaki Wakatsuchi1

Received 13 March 2001; revised 3 November 2002; accepted 5 February 2003; published 13 June 2003.

[1] Russian historical data and recently completed conductivity-temperature-depthsurveys are used to examine the formation and spread in the deep Ohkotsk Sea of denseshelf water (DSW) produced in the Okhotsk Sea polynyas. Isopycnal analysis indicatesthat all of the main polynyas contribute to the ventilation at sq < 26.80, including theOkhotsk Sea Mode Water (OSMW), which has densities sq = 26.7–27.0. At densitiesgreater than 26.9 sq the northwest polynya is the only contributor to OSMW. (AlthoughShelikhov Bay polynyas produce the densest water with sq > 27.1, vigorous tidal mixingleads to outflow of water with a density of only about 26.7 sq). In the western OkhotskSea the East Sakhalin Current rapidly transports modified dense shelf water along theeastern Sakhalin slope to the Kuril Basin, where it is subject to further mixing because ofthe large anticyclonic eddies and tides. Most of the dense water flows off the shelves inspring. Their average flux does not exceed 0.2 Sv in summer and fall. The shelf watertransport and water exchange with the North Pacific cause large seasonal variations oftemperature at densities of 26.7–27.0 sq (depths of 150–500 m) in the Kuril Basin,where the average temperature minimum occurs in April–May, and the averagetemperature maximum occurs in September, with a range of 0.2�–0.7�C. The averageseasonal variations of salinity are quite small and do not exceed 0.05 psu. The SoyaWater mixed by winter convection, penetrating to depths greater than 200 m, in thesouthern Kuril Basin also produces freezing water with density greater than 26.7 sq.Using a simple isopycnal box model and seasonal observations, the OSMW productionrate is seen to increase in summer up to 2.2 ± 1.7 Sv, mainly because of increased NorthPacific inflow, and drops in winter to 0.2 ± 0.1 Sv. A compensating decrease intemperature in the Kuril Basin implies a DSW volume transport of 1.4 ± 1.1 Sv fromFebruary through May. The residence time of the OSMW in the Kuril Basin is 2 ± 1.7years. INDEX TERMS: 4223 Oceanography: General: Descriptive and regional oceanography; 4243

Oceanography: General: Marginal and semienclosed seas; 4283 Oceanography: General: Water masses; 4536

Oceanography: Physical: Hydrography; 4540 Oceanography: Physical: Ice mechanics and air/sea/ice

exchange processes; KEYWORDS: Okhotsk Sea, North Pacific Intermediate Water, Mode Water, water mass,

marginal sea

Citation: Gladyshev, S., L. Talley, G. Kantakov, G. Khen, and M. Wakatsuchi, Distribution, formation, and seasonal variability of

Okhotsk Sea Mode Water, J. Geophys. Res., 108(C6), 3186, doi:10.1029/2001JC000877, 2003.

1. Introduction

[2] The Okhotsk Sea (Figure 1) is a marginal semien-closed sea that is the main source of cold, less saline, andoxygenated intermediate water in the western North Pacific

[Moroshkin, 1966; Kitani, 1973; Favorite et al., 1976;Talley, 1991, 1993; Talley and Nagata, 1995; Yasuda,1997; Watanabe and Wakatsuchi, 1998; Freeland et al.,1998; Wong et al., 1998; Itoh, 2000]. There are no directobservations of deep dense convection penetrating to inter-mediate density range (26.7–27.6 sq) in the open OkhotskSea. Talley [1991] showed that saline Soya Water (SW)entering in the southern Kuril Basin from the Japan Seacould be a source of dense convection, but also showed thatvigorous mixing with low-salinity Okhotsk water signifi-cantly reduces the maximum winter density in the KurilBasin. The principal source of ventilation at 26.7–27.0 sq isdense shelf water (DSW) produced by brine rejection incoastal polynyas during ice formation [Kitani, 1973; Alfultisand Martin, 1987; Talley, 1991; Martin et al., 1998;

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 108, NO. C6, 3186, doi:10.1029/2001JC000877, 2003

1Institute of Low Temperature Science, Hokkaido University, Sapporo,Japan.

2Scripps Institution of Oceanography, University of California, SanDiego, La Jolla, California, USA.

3Sakhalin Research Institute of Fisheries and Oceanography, Yuzhno-Sakhalinsk, Russia.

4Pacific Scientific Research Fisheries Center, Vladivostok, Russia.

Copyright 2003 by the American Geophysical Union.0148-0227/03/2001JC000877$09.00

17 - 1

Watanabe and Wakatsuchi, 1998; Wong et al., 1998; Glady-shev et al., 2000]. Water in this density range is the upperpart of the Okhotsk Intermediate Water. Yasuda [1997]called this Okhotsk Sea Mode Water (OSMW), because ofits thickness (low potential vorticity). We use a slightlymore restricted density range for OSMW than Yasuda(26.6–27.0 sq). The low potential vorticity of the OSMWreflects convection and mixing in the Kuril Basin but doesnot explain the strong ventilation of the entire intermediatelayer.[3] Generally, the OSMW is a mixture of three compo-

nents. The first and maybe largest is North Pacific Water(NPW) entering the Okhotsk Sea through the Kuril Straits.The quantitative contribution of this component to theOSMW is unknown and presents the biggest gap in ocean-ographic observations in this region. Repeat sections inBussol’ and Kruzenshtern Straits, the two largest KurilStraits, clearly show the dominance of tidal variability,rendering geostrophic transport estimates meaningless [Ris-er, 1996]. Recent theoretical estimates of tidally generatedtransport give a value of about 3.4 Sv of NPW flow into theOkhotsk Sea through Bussol’, Kruzenshtern, and FrizaStraits [Nakamura et al., 2000a]. It is not clear, however,what part of this transport occurs in the OSMW densityrange.

[4] The second component contributing to the OSMW isproduced in winter coastal polynyas. This is the coldest andleast saline component of the OSMW. Using satellite data,Alfultis and Martin [1987], Martin et al. [1998], and Glady-shev et al. [2000] showed that polynyas on the northernshelves start to form in January and end at the end of March.Using a relation between ice formation and brine rejection,the northwest shelf (NWS) polynya is always the main brinewater producer, contributing more than 50% of the totalproduction, and the Shelikhov Bay polynya is usually thesecond with about 25% of the total, regardless of largeinterannual variability. Recently, Gladyshev et al. [2000]analyzed extensive conductivity-temperature-depth (CTD)surveys on the northern Okhotsk shelves and found waterwith density greater than 26.9 sq and close to the freezingpoint on the NWS and in Shelikhov Bay in 1996–1997. Agravity current was hypothesized to transport this DSW toSakhalin Bay from the NWS, generating cyclonic circula-tion over the northern shelf, while Shelikhov dense waterdrained directly to the Tinro Basin. In 1996–1997 theestimated annually averaged volume flux of the DSW was0.5 and 0.24 Sv at densities 26.6–26.9 sq. However, mostof this water left the shelves before July, with a volumetransport of about 0.8 Sv, and further flux did not exceed 0.2Sv. A slow offshore DSW flux during the second half of the

Figure 1. Chart of the Okhotsk Sea. The 200 m depth contour defines the shelves. The abbreviationsare Hok for Hokkaido, Sak for Sakhalin, Skb for Sakhalin Bay, NWS for northwest shelf, NS for northernshelf, Kp for Koni-Pyagina, Shl for Shelikhov, and Kam for Kamchatka. Sections along L1, L2, and L3are shown in Figures 5 and 11.

17 - 2 GLADYSHEV ET AL.: OKHOTSK SEA MODE WATER

year and cyclonic circulation over the northern shelvesexplain the existence of DSW on the NWS even in Sep-tember. The general scheme of DSW expansion in the deepOkhotsk Sea was proposed by Kitani [1973] and had beenconfirmed by a number of other investigators. Kitanishowed that DSW flows off the continental shelf andspreads southward along the western Okhotsk Sea margin,diffusing into the OSMW through enhanced horizontalmixing in the central Okhotsk Sea.[5] The third component of the OSMW is the Soya Water

(SW) [Takizawa, 1982; Talley, 1991; Watanabe and Wakat-suchi, 1998; Itoh, 2000]. This is the warmest and saltiestcomponent of the OSMW, flowing through the narrow(about 42 km) and shallow (about 55 m) Soya Strait [Talleyand Nagata, 1995]. Watanabe and Wakatsuchi [1998]referred to the OSMW as Kuril Basin Intermediate Water.They showed that the Forerunner SW is an important sourceof dissolved oxygen for OSMW in the Kuril Basin. The SWtransport into the Okhotsk Sea varies from about 0.1 Sv inDecember–February to about 1.2 Sv in June [Takizawa,1982; after Aota, 1975].[6] The OSMW differs from the North Pacific Water

(NPW) in the same density range [Moroshkin, 1966;Yasuda, 1997]. OSMW is significantly thicker by about250 m than NPW and has been identified as a pycnostad orlow potential vorticity water [Yasuda, 1997]. Talley [1993]attributed the low potential vorticity of OSMW entering theNorth Pacific to relative vorticity, but Okhotsk Sea convec-tion and mixing is the most likely origin of the low potentialvorticity. According to Moroshkin [1966], who referred tothe OSMW as the Transitional Water Mass, its greatestthickness is 450–650 m in the Kuril Basin (between 150–200 m and 600–800 m). Its thickness decreases northwardto about 250 m (between 150–200 m and 300–400 m).Moroshkin [1966] explained the deepening of OSMW lowerboundary in the Kuril Basin by intensive downwelling. TheOSMW is characterized by temperature 0.1�–2�C andsalinity 33.3–33.8 psu. Oxygen saturation decreases from60–70% (6.5 mL/L) at the upper OSMW boundary to 15–20% (2.5 mL/L) at the lower OSMW boundary. However,oxygen saturation can exceed 70% in the western OSMW[Moroshkin, 1966].[7] Gladyshev et al. [2000] showed that the estimated

DSW outflow implies a residence time of OSMW at 26.7–26.8 sq of about 4 years, and at 26.8–26.9 sq of about 14years. However, Wong et al. [1998], using chlorofluorocar-bon data, found that the OSMW, which they referred to asupper Okhotsk Sea Intermediate Water, is renewed within1.4 years, with a correspondingly larger estimated DSWflux. The disagreement in residence time estimates indicatesthat further work is necessary.[8] Although winter convection occurs in the upper 100–

150 m layer in the central Okhotsk Sea and in the 300–400m layer in Kuril Straits according to Moroshkin [1966],Kitani [1973] reported that seasonal variations of thermo-haline structure extend to 500 m depth in the Okhotsk Sea.This depth corresponds to a density of about 27.0 sq, that is,the maximum density of the DSW on the NWS; thisconsistency is intriguing. On the other hand using fiveoceanographic surveys obtained in summer–fall during1977 through 1979, Wakatsuchi and Martin [1991, Figure6] showed large seasonal changes in the upper 1200 m of

the anticyclonic eddy located in the eastern Kuril Basin.They explained the variations by summer advection of cold,less saline and oxygen-rich water from Terpeniya Bay andthe east Sakhalin shelf.[9] In this paper we examine the DSW formation on the

shelves, its further spread toward the deep Okhotsk Sea, andrelated seasonal OSMW variability in the Kuril Basin. Theresidence time of the OSMW is estimated on the basis ofthese seasonal changes. Section 2 briefly describes our dataset. Section 3 shows the sources of dense cold water in theOkhotsk Sea contributing to the OSMW formation. Section4 describes the distribution of the average temperature andsalinity on isopycnals using the historical data and synopticsurveys. Section 5 discusses the seasonal changes on thenorthern shelves and in the OSMW, in the DSW and NPWfluxes, and in the OSMW production rate, using a simpleisopycnal box model. The residence time of OSMW is alsoestimated.[10] Acronyms are listed in Table 1.

2. Data

[11] The data set consists of historical bottle data takenbetween 1930 and 1995 and archived at Pacific Oceano-logical Institute in Vladivostok, Pacific Scientific ResearchFisheries Center in Vladivostok, Research Institute of Fish-eries and Oceanography in Yuzhno-Sakalinsk, Russia aswell as the recently completed CTD surveys (Table 2)whose results are being published in more detail elsewhere.Extensive sea ice in winter reduces the number of stationsbetween December and March. The CTD data have 1 dbarvertical resolution, and temperature and salinity accuracy of0.01� C and 0.01 psu or better. The salinities were calibratedwith water bottle samples. Cruises marked by asterisks inTable 2 are at WOCE standard quality [Freeland et al.,1998; Riser et al., 1996].[12] For bottle data quality control and processing, we

followed the procedure described by Boyer and Levitus[1994]. Quality control was made separately for the deepand shelf parts of the Okhotsk Sea, divided by an isobath of200 m. In the deep part, data were averaged for statisticalchecks by 2� squares. In the shelf region, we distinguish theHokkaido, Sakhalin, Sakhalin Bay, Northwestern Shelf(NWS), Northern Shelf (NS), Koni-Pyagin shelf (KP),Shelikhov Bay (SHL), and Kamchatka shelf (KAM) (Figure1), where the stations were sorted into depth bins between 0and 25 m, 25–50 m, 50–100 m, 100–150 m, and 150–200m within each 2� box. Following the statistical check, thehistorical data were averaged by 1� squares in the deep part

Table 1. Acronyms

Acronym Definition

DCW Dense Cold WaterDSW Dense Shelf WaterMDSW Modified DSWMNPW Modified NPWNPW North Pacific WaterOSMW Okhotsk Sea Mode WaterSW Soya WaterESC East Sakhalin CurrentNS Northern shelfNWS Northwest shelf

GLADYSHEV ET AL.: OKHOTSK SEA MODE WATER 17 - 3

and by the given depth bins within 1� squares on the shelvesfor input to objective analysis. For the distribution ofbottom density on the shelves, we used only measurementstaken within 20 m of the bottom.[13] After quality control, we chose 20,447 hydrographic

stations, in addition to the 1840 CTD data stations (Figure 2).The Okhotsk Sea is well sampled except in PenginskayaBay, the northeasternmost part of the Shelikhov Bay, andthe central Okhotsk Sea (50–54.5�N and 146�–149�E).Sampling is denser in the shelf areas. 7529 stations are

included for the shallowest analyzed surface, 26.7 sq,decreases to 3341 on the densest OSMW surface, 27.0 sq.The temporal distribution of historical data (Figure 3)reflects a weak exploration of the Okhotsk Sea before WorldWar II, with a rapid increase in the 1950s during the Vityazexpeditions, which were the observational basis forMorosh-kin [1966]. Field measurements were intensive in the 1960–1970s, with more than 200 stations per year. Much lowercoverage from the late 1970s to late 1980s may reflectincompleteness of our data set.

Table 2. CTD Data Summary

Ship Date Number of Stations Institution

Akademik Lavrentyev September 1989 11 POIAkademik Nesmeyanova [Freeland et al., 1998] September 1993 33 POI-IOSAkademik Lavrentyev July–Aug. 1994 135 POIPulkovo March–April 1995 119 PSRFCAkademik Lavrentyeva [Riser et al., 1996] April –May 1995 155 POI-UW-SIOProfessor Levanidov Nov.–Dec 1995 116 PSRFCTINRO April –June 1996 355 PSRFCProfessor Gagarinskii June–July 1996 117 POI-RIFOTINRO March–June 1997 393 PSRFCProfessor Levanidov July–Aug. 1997 137 PCRFCTINRO Aug–Sept. 1997 126 PSRFCProfessor Khromova July–Aug. 1998 64 Hydromet, HU, UWProfessor Khromova Aug–Sept. 1999 79 Hydromet, HU, SIO

aWOCE standard quality cruises. The abbreviations are Hydromet for Far-Eastern Hydrometeorological Research Institute (Russia), IOS for Institute ofOcean Sciences (Canada), HU for Low Temperature Science Institute, Hokkaido University (Japan), PSRFC for Pacific Scientific Research FisheriesCenter (Russia), POI for Pacific Oceanological Institute (Russia), RIFO for Research Institute of Fisheries and Oceanography (Russia), SIO for ScrippsInstitution of Oceanography (USA), UW for School of Oceanography, University of Washington (USA).

Figure 2. The distribution of historical data in the Okhotsk Sea including CTD cruises listed in Table 2.


[14] Potential temperature and potential density referencedto the sea surface are used throughout the following text.The words ‘‘temperature’’ and ‘‘density’’ mean ‘‘potentialtemperature’’ and ‘‘potential density’’.

3. Sources of Dense Cold Water in the OkhotskSea

[15] Talley [1991] discussed possible sources of densecold water (DCW) in the Okhotsk Sea using data from theWorld Data Center, most of Japanese origin. We use ourexpanded data set to clarify the dominant processes. Wedefine the DCW and DSW as water with temperature below0�C and density greater than 26.7 sq. However, in contrastto the DSW, the DCW can form not only on the shelves butalso in the open Okhotsk Sea and, hence this term is moregeneral. Winter convection and brine rejection during iceformation in polynyas [Alfultis and Martin, 1987; Martin etal., 1998] are considered including the contributions of theShelikhov Bay and Sakhalin polynyas. Gladyshev et al.[2000] described DSW production on the NS and NWS, soour discussion of this region is confined to section 5, wherethe seasonal variability of OSMW is considered.

3.1. Potential for Winter Convection in the DeepOkhotsk Sea

[16] Since ice cover has prevented direct observations ofwinter convection in the western and central Okhotsk Seauntil very recently, investigators have used indirect meth-ods, such as density of the temperature minimum in thedichothermal layer [Kitani, 1973], to infer the density ofconvection. Talley [1991], using late winter and spring datain the ice-free regions, found 26.7 sq outcropping close tothe northern Kuril Islands and at Soya Strait, where thesurface densities exceeded 26.8 sq. The latter is due toinflow of the spring Soya Warm Current (SWC) [Takizawa,

1982; Talley and Nagata, 1995; Watanabe and Wakatsuchi,1998; Itoh, 2000]. Spring (‘‘Forerunner’’) SW [Takizawa,1982] spreads along the Hokkaido coast between March andMay with a 0.2 Sv transport concentrated at 26.8–26.9 sq[Watanabe and Wakatsuchi, 1998]. Although the spring SWis important for water mass modification in the Kuril Basin,its temperature exceeds 2�C [Takizawa, 1982] and itstransport is small.[17] The Soya Warm Current flows from March through

November with a considerably larger volume transport ofabout 1 Sv in summer–fall [Takizawa, 1982; Talley andNagata, 1995; Itoh, 2000]. Consequently, the summer–fallcontribution of the SW to the southern part of the OkhotskSea is greater. The summer SW salinity is greater than 33.6psu, but its density is less than 26.3 sq because of hightemperature and spreads in the upper layer in the KurilBasin. The behavior of the SW after passing the tip of CapeShiretoko is not clear [Talley and Nagata, 1995], and itprobably mixes through eddies into the Kuril Basin [Taki-zawa, 1982; Bobkov, 1993]. SW outflow to the NorthPacific was observed through Friza Strait in August 1989.The saline SW is a tantalizing potential source of dense coldwater in winter. However, Talley [1991] hypothesized, onthe basis of limited winter surface data in the Kuril Basin,that SW mixes with less saline Okhotsk Sea water beforecooling, reducing the density of the convected surface waterto less than 26.8 sq.[18] To examine winter convection in the deep Okhotsk

Sea, we use the Prof. Levanidov November–December1995 cruise (Table 2), when winter convection had alreadybegun in the upper 50–80 m. The September Ak. Nesmeya-nov 1993 and Prof. Khromov 1999 data were used to fill adata gap in the Deryugin Basin, where fresh Amur Riverwater dominates the surface salinity field (Figure 4a). Thereare two regions where surface salinity exceeds 33.0 psu andso the density would exceed 26.6 sq if cooled to the freezingpoint. The first is Bussol’ Strait in the central Kuril Islandswhere strong tidal mixing in the Kuril Straits brings higher-salinity water to the surface [Moroshkin, 1966; Kowalik andPolyakov, 1998; Nakamura et al., 2000b]. The secondregion is the Southern Kuril Basin where SWC is widelydistributed in fall [Itoh, 2000] and actively mixes with low-salinity water from the East Sakhalin Current (ESC). A low-salinity tongue intrudes southeastward to Iturup (Etorofu)Island and almost splits the surface SW in two. Most of theOkhotsk Sea surface salinity is between 32.4 and 32.9 psu atthe beginning of winter, with lowest salinity in the westwhere the ESC transports the Amur River water alongSakhalin to the Hokkaido coast, where it is often observedin November [Watanabe, 1963].[19] We estimate the density of potential winter convec-

tion using this December salinity distribution, assuming thatsurface salinity does not change in winter and that upperlayer density varies only because of local convection. Wealso assume for the calculation that ice-covered areas are allat the freezing point, although some of the Okhotsk Sea icecover, particularly at the southern edges, is exported fromice production areas and is melting. Thus the estimateddensities are the maximum possible and are likely high,particularly along the southern edge of the ice region. First,the maximum density of the convectively mixed layer iscalculated for each station as rc = f(Tw, Ssr), where Ssr is the

Figure 3. The year-to-year distribution of the historicaldata.


surface salinity (Figure 4a) and Tw is the winter surfacetemperature. The latter is taken at the freezing point for thegiven salinity if the station is covered by ice in winter, or as0�C for open water in accord with winter observations[Talley and Nagata, 1995] (Figure 4a). The mean OkhotskSea ice extent for 1971–1990 [Talley and Nagata, 1995]was used. The depth of rc on each actual vertical profile is afirst iteration of the mixed layer thickness. Then the averagesalinity is calculated in the estimated mixed layer and rc iscalculated again with the new salinity to produce anestimate of the density and thickness of winter convection(Figures 4b and 4c). (Although ice formation increases thesalinity (density) of the mixed layer, we do not include thiseffect. For example, formation of 0.5 m ice, which is typicalfor the central Okhotsk Sea, will increase salinity of a 50 m(100 m) layer by about 0.3 (0.15) psu.)[20] The estimated convection is relatively shallow in the

northern and central Okhotsk Sea, with a thickness of 50–75 m and densities 26.2–26.4 sq that could be 0.1–0.2 sqhigher depending on ice formation. This is consistent withnumerous summer observations of depth and thickness ofthe dichothermal layer, which is generally shallower than100 m and with a density of about 26.5 sq (not shown). Theestimated convection is shallowest along the Sakhalin shelf,with the lowest density of about 24–25 sq. Note howeverthat September data are used in the northwestern OkhotskSea, and hence our estimates may be most uncertain for thisregion. It is evident that the low-salinity surface water

derived from the Amur River could inhibit DCW formationin the northern and central Okhotsk Sea.[21] In contrast to the northern area, estimated winter

convection could reach to greater than 200 m in the KurilStraits and in the region adjacent to the southern KurilIslands (Figure 4c). However, this estimated depth ismeaningless in the straits, as surface temperatures remainwell above freezing all winter. The higher salinity here isdue to strong tidal mixing with the warmer, saltier NorthPacific water, so the high salinity cannot create favorableconditions for convection.[22] The relatively dense estimated winter convection in

the southern Kuril Basin is caused by the saline SW (Figure4b). Mixing of the SW with the low-salinity ESC water inthe Kuril Basin’s vigorous eddy field causes large spatialvariations of the depth and density of the dichothermallayer. A Kuril Basin section from May 1995 (Figure 5a)shows two different dichothermal layers. In the regionadjacent to Sakhalin, the dichothermal layer is shallow(<100 m) and low density (26.4–26.5 sq) because thewinter convection occurred in the low-salinity surfacewater. The dichothermal layer adjacent to Iturup has adensity of 26.6–26.7 sq and a depth of 150–250 m becauseof saline SW, spreading in this region in the previous fall(not shown). High dissolved oxygen and low dissolvedsilicate at 26.7 sq in the vicinity of the Iturup (not shown)are also consistent with an hypothesis of the local formationof this water.

Figure 4. (a) Surface salinity, (b) surface density, and (c) mixed layer depth based on the data ofNovember–December Prof. Levanidov 1995 cruise (solid circles) and September Ak. Nesmeyanov 1993and Prof. Khromov 1999 cruises (open circles). The dashed line shows the long-term mean ice boundaryin March. Shaded areas encompass the regions where salinity (Figure 4a) and density (Figure 4b) exceed33 psu and 26.6 sq, correspondingly. The cross shaded area denotes the DCW region in April–May 1995.The contour interval is 0.5 for salinity 30–32 psu and 0.2 for salinity greater than 32 psu. The contourinterval is 0.5 for density 24–26 sq and 0.1 for density greater than 26.0 sq. The thin line is the 300 misobath.


[23] The location of DCW in April–May 1995 is shownin Figure 4c. On the basis of the direct CTD measurements,its thickness exceeds 30 m near Iturup and is greater than 50m at the southernmost station (44.99�N, 146.59�E). It islikely that the convective DCW forms in February–March,when ice extends to the southeastern Kuril Basin [Talley,1991; Wakatsuchi and Martin, 1991; Yasuda, 1997]. Usingthe Lavrentiev April–May 1995 cruise, we estimate avolume of the DCW produced by the winter convection inthe Kuril Basin of about 500 km3, which gives its Febru-ary–March production rate of about 0.1 Sv. We note thatthis is a very crude estimate because there are few winterobservations south of 45�N in the Kuril Basin. Although theDCW production due to convection is apparently low incomparison with the DSW production on the northernshelves [Gladyshev et al., 2000], we need a more accurateestimate using reliable data. We have to caution that it isdifficult to separate the DCW from the DSW in the KurilBasin using only temperature and salinity. We are not able

to do this employing our historical data, although we findwater with temperature below 0�C at 26.7 sq along thesouthern Kuril Islands in spring. Hydrochemical tracerswould be helpful, keeping in mind that the advectiveDSW loses its properties because of mixing en route tothe Kuril Basin.

3.2. Shelikhov Bay

[24] Shelikhov Bay, in the northeastern Okhotsk Sea, is asite of dense (>26.9 sq) shelf water formation through iceproduction in winter [Gladyshev et al., 2000]. The DSWextends southward to a neck of Shelikhov Bay and coulddrain directly to the Tinro Basin. However, the two avail-able cruises showed only lower densities (<26.7 sq) outsidethe Shelikhov entrance, from which we conclude thatShelikhov Bay does not really contribute to the OSMWformation, as follows.[25] Bottom temperature, salinity, and density in

March–May 1997 and bottom oxygen for July–August1997 reveals two separate areas where the bottomdensity exceeds 26.7 sq, with considerably differentwater properties. The northern area includes the Sheli-khov DSW [Gladyshev et al., 2000], with q < 0� and S> 33.2 psu. The southern, much larger area is filled byOSMW with q > 1�C and S > 33.3 psu. In the Shelikhovneck, bottom salinity and density are less than 32.9 psuand 26.5 sq, presumably because of vigorous tides[Kowalik and Polyakov, 1998]. Water of the same prop-erties as in the neck is widely distributed over the bottomof Koni-Pyagin shelf (Figure 6d) to the west of PyaginPeninsula. There is no sign of DSW on the continentalslope of the Koni-Pyagin shelf and in the Tinro Basin inMarch–May 1997 (Figure 6). These indicate that waterexiting Shelikov Bay flows westward along the shelf inaccordance with the general cyclonic circulation in theOkhotsk Sea.[26] In Tinro Basin, oxygen is low at the bottom (Figure

6c). Moreover, at 26.9 sq oxygen is lower (<80 mM/kg)here than elsewhere in the Okhotsk Sea, while silicate(>150 mM/kg) is the highest (not shown). At depths greaterthan 600 m, Sagalevich and Ivanenkov [1982] found, fromsubmarine observations, extremely low visibility, of 2–3m, and very high turbidity due to the high concentration oforganic matter, up to 100–200 mg/L. This was associatedwith high biological productivity of the western Kam-chatka shelf. The organic matter sinks to the bottom,producing 6 km thick layer of deposits, which is thethickest in the Okhotsk Sea. This suggests a large amountof remineralization of organic matter in the Tinro Basin. Itseems likely that this process also occurs in the DeryuginBasin, where very high (low) silicate (oxygen) has beenobserved at the bottom [Freeland et al., 1998]. In contrastto Tinro Basin, oxygen in the tidally mixed bottom waterin Shelikhov Bay is highest in summer (Figure 6c). Theenormously large tides completely mix the DSW in She-likhov Bay between May and August [Gladyshev et al.,2000].

3.3. Sakhalin Shelf

[27] The Sakhalin and Terpeniya Bay polynyas maycontribute slightly to the DSW production according toMartin et al. [1998] using Chapman and Gawarkiewicz’s

Figure 5. (a) The distribution of potential temperature(solid lines) and potential density (dashed lines) on thevertical section crossing the Kuril Basin between Sakhalinand Iturup Islands. The location of the section is shown inFigure 1 as L1. (b) The same for a section at 55�N from theshelf north of Sakhalin eastward to the central Okhotsk Sea(section L2 in Figure 1). Temperature below zero is shaded.


[1997] theoretical results. All available in situ data forApril–June (Figure 7) are used here to examine thispotential contribution. Ice cover over the Sakhalin shelfnorth of Terpeniya Cape continues into May [Martin et al,1998; Talley and Nagata, 1995], so these data representthe earliest period after ice melt. DSW is found only inSakhalin Bay and on the eastern Sakhalin shelf north of53�N (Figure 7), and originates from the northern shelves[Gladyshev et al., 2000]. This dense cold water at northernSakhalin is illustrated in Figure 5b, which is a section (L2 inFigure 1) along 55�N. The rest of the Sakhalin shelf has noDSW, excluding the narrow area adjacent to the 200 m

isobath where it is nevertheless difficult to identify theorigin of water of density >26.7 sq.[28] There are at least two reasons why DSW is not

found on the Sakhalin shelf even just after ice melt. First,the eastern Sakhalin shelf is only 20–50 km wide and theDSW, which is produced before the end of March[Martin et al., 1998; Gladyshev et al., 2000], could beremoved from the shelf by currents before May–June.Also, strong tidal currents [Kowalik and Polyakov, 1998]intensively mix the DSW in the coastal area shallowerthan 50 m [Moroshkin, 1966]. Second, the Amur Riverfresh water dominates the surface in the western Okhotsk

Figure 6. Bottom (a) potential temperature (�C), (b) salinity, (c) dissolved oxygen (mM/kg), and (d)potential density (kg m�3) in Shelikhov Bay and adjacent area. Station locations: from March to May1997 (Figures 6a, 6b, and 6d) and from July to August 1997 (Figure 6c). Shading indicates potentialdensity exceeding 26.7 sq. Cross hatching indicates potential density exceeding 26.9 sq. The thin lineindicates the 200 m isobath.


Sea as is evident in the sharp surface salinity frontformed in June–July 1996 near Sakhalin (Figure 8).The Amur runoff spreads in Sakhalin Bay over thenorthwestern escarpment of the Deryugin Basin and overthe eastern Sakhalin shelf north of 50�N. Further, theESC transports the low-salinity water southward along theshelf, causing its appearance at the Hokkaido coast inNovember [Watanabe, 1963]. When convectively mixedin winter with shelf bottom water, the fresh surface waterconsiderably decreases the water column salinity, suchthat the brine rejection in polynyas may not be able toproduce DSW.[29] Winter density on the Sakhalin shelf can be inferred

from the June–July 1996 data, assuming that water shal-lower than the 150 m isobath completely mixes by winterconvection [Moroshkin, 1966] (Figure 8b). For the openOkhotsk Sea we assume that the temperature minimum isthe lower boundary of winter convection. This ‘‘winter’’density varies significantly from 25.5 sq north of 52�N to26.4 sq south of 50�N on the Sakhalin shelf (Figure 8b),hence not dense enough to become DSW. Martin et al.[1998], using Chapman and Gawarkiewicz’s [1997] model,estimated that brine rejection in Sakhalin polynyas increasesthe bottom density by 0.3 sq. Hence the DSW formationdensity should be about 26.7 sq The isopycnal analysis insection 4 will shed some light on this problem. Interannualvariability in Amur River runoff may be a controlling factor

for DSW formation on the Sakhalin shelf which can changeby a factor of two [Rogachev, 2000].

4. Basin-Wide Okhotsk Sea Mode WaterProperties

[30] To describe the spatial distribution of the OSMW inthe open sea together with the DSW on the shelves, andthe density to which the DSW can increase because ofbrine rejection in polynyas, we first present isopycnalmaps of isopycnally averaged potential temperature andsalinity using all available data. It is easy to identify theDSW on these maps, because its temperature is well below0�C on the shelf all year [Gladyshev et al., 2000] as isalso apparent in vertical cross sections from the shelf(Figure 5b). However, the freezing DSW rapidly losesits extreme properties near the shelf break as it descendsfrom the shelves, assumed primarily because of tidalmixing [Kowalik and Polyakov, 1998]. Tidal mixing alsomodifies warm and saline NPW entering through the Kurilstraits. You et al. [2000] show that diffusive doublediffusion also contributes to the transformation of OSMWfrom 26.8 to 27.4 sq.[31] In order to trace the influence of DSWand NPWafter

they have been modified and to differentiate this modifiedDSW (MDSW) and modified NPW (MNPW) from theOSMW in the deep Okhotsk Sea, we define MDSW and

Figure 7. Mean bottom (a) potential temperature, (b) salinity, and (c) potential density on the easternSakhalin shelf in April–June based on the historical data. Shading indicates density greater than 26.7 sq.The thin line is the 200 m isobath.


MNPW as follows. (MDSW and MNPW are not tradition-ally defined water masses, but are useful for tracking themixed products of DSW and NPW.) MDSW is colder andfresher than OSMW, while MNPW is warmer and saltier. AMDSW (MNPW) upper (lower) bound separately for eachisopycnal surface from 26.7 to 27.0 sq is defined as

quMDSW ¼ �qOSMW � 3*s; qlMNPW ¼ �qOSMW þ 3*s; ð1Þ

and

SuMDSW ¼ �SOSMW � 3*s; SlMNPW ¼ �SOSMW þ 3*s;

where qMDSWu and SMDSW

u are the upper thermohaline boundof the MDSW, qMNPW

l and SMNPWl are the lower thermoha-

line bound of the MNPW, and �qOSMW and �SOSMW are theisopycnally averaged temperature and salinity of OSMW,and s is the standard deviation of mean of the OSMWproperties (Table 3). The use of 3s to define the boundaries

is arbitrary but useful. The MDSW and MNPW boundariesare shown only on temperature maps (Figure 9). Theboundaries based on salinity are consistent with thetemperature boundaries. However, the OSMW salinityrange on isopycnal surfaces is small and three standarddeviations in salinity just slightly exceed the accuracy of thehistorical data, which is 0.02 psu. The ‘‘MNPW boundary’’in the southern Okhotsk Sea along the Hokkaido coastactually shows Forerunner SW. The coldest OSMW is mostrecently ventilated by the DSW.

4.1. Okhotsk Sea Mode Water Using All AvailableData

[32] At 26.7 sq, cold DSW (q < �1�C, S < 33.20 psu) iswidely distributed on the NS, NWS, in Sakhalin Bay whereit is transported from the NWS, and in Shelikhov Bay(Figures 9a and 9b), consistent with Kitani [1973] andGladyshev et al. [2000]. The DSW rapidly changes itsproperties in the neck of Shelikhov Bay (Section 3). Alongthe eastern Sakhalin shelf the DSW descends and forms a

Figure 8. Surface (a) salinity in June–July 1996 and (b) calculated density of winter convection in thewestern Okhotsk Sea. The dots show the station locations. The contour interval is 5.0 for salinity 20–30psu, 1.0 for salinity 30–31 psu, and 0.1 for salinity greater than 31 psu. The contour interval is 0.5 fordensity 25–26 sq and 0.1 for density greater than 26.0 sq. The thin line is the 200 m isobath.

Table 3. Average Isopycnal Values for the OSMW (�qOSMW, �SOSMW) and the Upper and Lower Thermohaline

Boundaries of the Modified Dense Shelf Water (qMDSW, SMDSW) and Modified North Pacific Water (qMNPW, SMNPW)

Defined as in the Text

sq �qOSMW, �C �SOSMW, psu quMDSW, �C SuMDSW, psu qlMNPW, �C SlMNPW, psu

26.7 0.48 ± 0.15 33.29 ± 0.01 �0 �33.26 �1.0 �33.3226.8 0.91 ± 0.10 33.44 ± 0.01 �0.6 �33.41 �1.2 �33.4726.9 1.17 ± 0.10 33.59 ± 0.01 �0.9 �33.56 �1.5 �33.6227.0 1.48 ± 0.08 33.73 ± 0.01 �1.2 �33.70 �1.7 �33.76


large, cold pool, longitudinally oriented in the westernOkhotsk Sea, with potential temperature below 0�C andsalinity less than 33.26 psu. Further transformation of theMDSW along its cyclonic pathway includes gradual mixingwith the warmer and saltier MNPW in the central OkhotskSea. The latter, indicated by the 1�C isotherm, is widelydistributed east of 150�E from Friz Strait in the south to anarrow band along western Kamchatka shelf, up to 58�N in

the north. The MDSW mixing with the warm and salineForerunner SW occurs in the southern part of the OkhotskSea. The Forerunner SW dominates on the Hokkaido shelfin March–May [Takizawa, 1982; Watanabe and Wakatsu-chi, 1998] and, according to Figure 9a, spreads to the northalong the southern Kuril Islands to 45�N. It is not possibleto distinguish the convective DCW in the southern KurilBasin because of averaging.

Figure 9. The average distribution of potential temperature and salinity on isopycnals in the OkhotskSea: (a) and (b) 26.7 sq; (c) and (d) 26.8 sq; (e) and (f) 26.9 sq; (g) and (e) 27.0 sq. Shading indicates nowater with such potential density. The dashed (dot-dashed) line shows the boundary of the MDSW(MNPW). The thin lines are the 200 and 2000 m isobaths.


[33] At 26.7 sq, the basin-wide variations in temperatureand in salinity are about 2�C and less than 0.15 psu. There isno DSW formation and hence no freezing bottom water onthe Hokkaido, Koni-Pyagin and Kamchatka shelves becauseof the lack of polynyas.[34] DSW with a density of 26.8 sq forms on the NWS

and NS (Figures 9c and 9d). This water has salinity less than33.3 psu and is close to the freezing point. The DSW is alsoinvolved in cyclonic circulation over the northern shelf,turns around Sakhalin to reach the northeastern Sakhalinshelf, north of about 53�N, and cascades toward the

Deryugin Basin [Gladyshev et al., 2000]. On the otherhand, no DSW of this density is found on the easternSakhalin shelf between 49�N and 53�N. It is clear, however,that DSW with a density of 26.8 sq sometimes forms in theTerpeniya Bay polynya and flows southward along thesouthern part of the eastern Sakhalin shelf. At this density,water below 0.6�C with salinity less than 33.41 (Table 3)occurs in a narrow band along the eastern Sakhalin slope,between 45� and 55�N, clearly manifesting the main path-way of the MDSW in the open sea. This pathway reflectsthe East Sakhalin Current flowing along the entire Sakhalin

Figure 9. (continued)


slope and transporting the MDSW southward. In section 3,we showed that Shelikhov Bay does not contribute toventilation of the Okhotsk Sea at densities greater than26.8 sq. Fairly homogeneous, warm water in the northeast-ern Okhotsk Sea supports this conclusion.[35] At 26.8 sq, the MNPW penetrates far to the north to

the Tinro Basin and further to the Koni-Pyagin shelf. It alsoflows toward Kashevarov Bank. It appears that the MNPWmixes slightly with the MDSW along the eastern part of theNS continental slope. In the southern Okhotsk Sea, theHokkaido shelf and slope restrain the SW spreading (Fig-ures 9c and 9d). Generally at 26.8 sq, the range of salinity inthe open Okhotsk Sea, excluding a quite small SW region,just slightly exceeds 0.1 psu, although the temperature rangeis about 2�C.[36] Potential temperature and salinity at 26.9–27.0 sq

indicate that the only source of DSW in this density range isthe NWS polynya, located along the northwestern coastbetween Ayan and Okhotsk City (Figures 9e–9h). TheDSW from the NWS fills up the northwestern escarpmentof the Deruygin Basin and flows southward along theeastern Sakhalin slope to the Kuril Basin. MNPW stretchesnorthward from Bussol’ Strait along the Kuril Islands andthe western Kamchatka shelf. Forerunner SW on the Hok-kaido shelf and continental slope is the warmest and saltiestcomponent of the OSMW, consistent with Takizawa [1982],Talley [1991], Watanabe and Wakatsuchi [1998], and Itoh[2000]. Shelikhov Bay DSW can reach a density of 27.1 sq(not shown). However, as mentioned before, this densewater does not ventilate the Okhotsk Sea although Sheli-khov Bay polynyas may be important for the Okhotsk Seaas ice producers.

4.2. Modified Dense Shelf Water Using Synoptic Data

[37] Detailed study of the MDSW distribution in the openOkhotsk Sea requires quasi-synoptic surveys because itsmain pathway off the shelves is the narrow region along theeastern Sakhalin slope. We chose 1996 for its good spatialstation coverage in the western Okhotsk Sea, and becausethere is detailed analysis of the DSW formation on thenorthern shelves during this year [Gladyshev et al., 2000].Data are combined from two cruises: TINRO (the westernpart of the survey) and Prof. Gagarinskii, both duringJune–July 1996 (Table 2). The thickness of the layer withq < 0�C and sq < 26.7 for each station of the combinedsurvey was calculated (Figure 10a). In 1996, the DSW,originating from the NWS, descended from the shelf towardthe northwestern Deryugin Basin south of 56�N, where itsthickness was greater than 50 m. A DSW cascade alsooccurred across the shelf break of Sakhalin Bay and acrossthe shelf break of the northeastern Sakhalin shelf, north of53�N. The ESC then transported the thick lenses of MDSWsouthward along the eastern Sakhalin slope parallel to theisobaths. The length of these lenses varied between 50 and160 km and their width was about 20–50 km. Theirthickness was 50–200 m, which was greater than upstreamat their source, implying active vertical mixing, most likelydue to tides. Some MDSW expansion eastward at 51�–53�N gives the impression that part of this water is involved inthe cyclonic circulation over the Deryugin Basin. Flowingto the south, the MDSW enveloped a Kuril Basin anticy-clonic eddy, which contained no DSW.

[38] Temperature and salinity at 26.7, 26.8, and 26.9 sq(Figures 10b–10g) are generally consistent with the MDSWthickness distribution (Figure 10a) and the historical dataanalysis (Figure 9). We chose these isopycnals becausewater in this density range formed on the northern shelvesin 1996 [Gladyshev et al., 2000]. In Figures 10b–10d, wealso plot the MDSWand MNPW boundaries for comparison(Table 3). As in the historical data, DSW colder than�1.5�C occurred only on the shelves in 1996. At 26.7 sq,the DSW left the shelf in the large region across theSakhalin shelf break. Consequently, a large cold pool withpotential temperature below 0�C existed in the westernOkhotsk Sea, where some MDSW was involved in thecyclonic circulation over the Deryugin Basin. A largeamount of MDSW occurred in the western Kuril Basin,where it may have coalesced with convectively formedDCW (relatively thick and cold lens close to Iturup Islandin Figure 10b).[39] At 26.8 sq, the DSW cascaded across the Sakhalin

Bay shelf break and northeastern Sakhalin shelf break. Thedistribution of MDSW in the western Okhotsk Sea wasquite similar to that at 26.7 sq.[40] In contrast, at 26.9 sq(Figure 10d), the MDSW was

not as abundant in the western Okhotsk Sea, consistent withGladyshev et al. [2000]. In 1996, the annually averagedproduction rate of DSW at 26.8–26.9 sq was only 0.04 Svor four times less than for 26.7–26.8 sq [Gladyshev et al.,2000]. The MDSW flow at 26.9 sq turned eastward nearTerpeniya Cape, spreading toward Bussol’ Strait in theKuril Basin. A few separated lenses of MDSW occurred.The MDSW was quite homogeneous in salinity at alldensities (Figures 10e–10g), with the lowest salinities alongthe Sakhalin slope, consistent with the historical dataanalysis.[41] MDSW can be clearly seen along 48.6�N from

Terpeniya Cape in June 1996 (Figure 11). (Sections throughthe whole of the Okhotsk Sea to Bussol’ Strait were shownby Freeland et al. [1998].) The MDSW spreads over theSakhalin slope in the upper 500 m layer, recognized by itsanomalously low temperature and salinity. A steep slope ofisopycnals at the shelf break and in the deeper layerssuggests the existence of a strong ESC. The DSW densitywas greater than 26.7 sq on the Sakhalin shelf and likelyoriginated from the Sakhalin polynya. The separate lens ofMDSW 100 km offshore suggests that the ESC is subject tothe influence of small eddies and other transients, contrib-uting to MDSW spreading and mixing.[42] To estimate a travel time necessary for the DSW

from its source to the Kuril Basin, geostrophic velocitiesrelative to 1000 dbar for June–July 1996 are calculated,recognizing that there is seasonal and interannual variationso the values given here are estimates only. Velocities at 300dbar correspond to a density of about 26.8 sq (Figure 12).Geostrophic velocities at intermediate depth are persistentlysouthward along the Sakhalin slope with speeds in excess of15 cm/s in the north and are the core of the ESC. This isconsistent with average current speed estimates based on the2 year long current measurements made over the Sakhalinslope at 53�N (G. Mizuta, personal communication, 2001).The average speed between 48�N and 54.5�N was about 10cm/s at 300 dbar and 15 cm/s at 200 dbar or 26.7 sq (notshown). Hence most MDSW from the 750 km between


northern Sakhalin and the Kuril Basin would pass within 2–3 months. For the MDSW at 26.9 sq, from the northernshelves, 4 months are required to reach the Kuril Basin at anaverage speed of 7 cm/s. According to Gladyshev et al.[2000], the first DSW flows from the NWS to Sakhalin Bay1.5–2 months after formation. Thus the DSW from theNWS would take 3.5–6 months (depending on density) toreach the Kuril Basin. Given that DSW formation started inmid-January in 1996 [Gladyshev et al., 2000], the firstMDSW could reach the Kuril Basin at the beginning of

May. For the DSW from the Sakhalin and Terpeniyapolynyas with densities of about 26.7 sq the travel time tothe Kuril Basin would not exceed 1–2 months. Hence thiswater could appear in the Kuril Basin as early as Marchalong with the DCW produced by winter convection.

5. Seasonal Variability of DSW and OSMW

[43] As shown in section 4, a large amount of the DSWproduced on the Okhotsk shelves at the beginning of each

Figure 10. (a) The thickness of the layer with water with q < 0�C and sq > 26.7 sq. Potentialtemperature and salinity at (b) and (e) 26.7 sq, (c) and (f ) at 26.8 sq, (d) and (g) at 26.9 sq for 4 June to 6July 1996 based on the combined CTD surveys. Regions of potential temperature below 0�C are shaded.The dots show the station locations. The thin lines show the 200 and 2000 m isobaths. The triangle is thecenter of the anticyclonic eddy in the Kuril Basin.


year is transported out along the eastern Sakhalin slope tothe Kuril Basin. This suggests examining the seasonalvariability associated with this outflow. In order to clarifythe response of the OSMW to the DSW flux, a detailedexamination of the seasonal variations on the northernshelves and in the Kuril Basin from the historical data isundertaken in this section.

5.1. Seasonal Variability on the Northern Shelves

[44] The NWS and NS polynyas are the biggest DSWproducers at 26.7–26.8 sq (Figure 9). Moreover, the NWSpolynya is the only producer at 26.9–27.0 sq, ventilatingthe OSMW. The Shelikhov Bay polynyas produce thedensest DSW in the Okhotsk Sea but contribute little toOSMW modification (Section 3). Thus we restrict our

further exploration of the shelves to the NS and NWS.We select and average shelf data over the bottom withinMay–June (spring) and September–October (fall) as rep-resentative months to show seasonal variability of the DSW(Figure 13).[45] In spring, DSW with density exceeding 26.7 sq

occupies the entire NWS, Sakhalin Bay and the NS westof 146�E. It appears that DSW with density greater than26.9 sq forms annually on the NWS, while on the NS theDSW usually has densities of 26.7–26.8 sq. The latter flowscyclonically over the shelf, gradually replacing the DSW onthe NWS during spring and summer [Gladyshev et al.,2000].[46] Bottom density changes in coastal regions exceed 1.0

sq between spring and fall (Figure 13). At depths shallower

Figure 10. (continued)


than 100 m, strong tides dramatically reduce the density ofthe bottom layer to less than 26.0 sq through mixing withsurface water which is seasonally warmed and diluted bymelted ice, numerous small rivers and precipitation. Theamount of river discharge and precipitation to some degreecontrols stratification at the beginning of the brine rejection.The river runoff maximum always occurs in spring anddepends on the winter snow pack, while precipitation isgreatest in summer and fall.[47] As seen in Figure 8, the Amur runoff does not

influence the NWS, which lies upstream, in contrast to

Sakhalin Bay and the Sakhalin shelf. This difference seemsto be crucial for formation of the densest water on the NWS.Small rivers entering the NWS bring only 57.2 km3 yr�1 offresh water [Rogachev, 2000]. Rain gauge data collected bythe National Climate Data Center (NCDC) show thatmaximum precipitation in the north of the Okhotsk Seaoccurs in Ayan (Figure 1). The annual long-term mean 84 ±24 cm yr�1 for the period 1932–1999 gives a fresh waterflux of about 65 km3 yr�1 for the entire NWS. Martin et al.[1998] and Gladyshev et al. [2000] estimated that annually,more than 100 km3 of sea ice forms in the NWS polynya.Most of sea ice, however, is thought to be removed from theNWS before melting because of northwestern winds[Kimura and Wakatsuchi, 1999]. Hence melting ice, thesmall river runoff, and precipitation are the main contrib-utors to the surface salinity variations on the NWS. Varia-tions in freshwater flux, along with winter air temperatureand surface wind, produce interannual variability of thedense water production in polynyas. A more detailedanalysis of this problem is under way.[48] Below 100 m on the shelves, seasonal variation of

DSW is much less, with DSW with sq > 26.7 in bothseasons on the NWS, Sakhalin Bay and NS west of 146�E(Figure 13). Although the density of the DSW decreases byabout 0.1 sq from spring to fall, likely because of outflowfrom the shelves and also likely because of tidal mixing,Figure 13 gives the impression that not much DSW leaves

Figure 11. (a) Potential temperature, (b) salinity, and (c)potential density on the vertical section along 48.6�Noffshore from Terpeniya Cape for 28 June 1996. Thelocation of section is shown in Figure 1 by L3. Temperaturebelow 0�C is shaded.

Figure 12. Geostrophic velocity at 300 db relative to 1000dbar in the region of the East Sakhalin Current for 4 June to6 July 1996 based on the combined CTD surveys. The thinline is the 200 m isobath.


the shelf over summer. To check this hypothesis, weestimate the average removal rate U of the DSW from theseregions between spring and fall, where we define U as thevolume change between 1 June and 1 October. To calculatethe volume, we multiply the bottom area encompassed by agiven isopycnal and limited seaward by an isobath of 200 mby the seasonally averaged thickness of the isopycnal layerderived from the historical data on the NWS and NS and inSakhalin Bay. Table 4 lists the average thickness (H) ofDSW at 26.7–26.9 sq, their volumes (V), and the summer–fall removal rates (U) with the estimated errors based onstandard deviation of mean.[49] As shown in Table 4, the summer–fall removal rate

for DSW with sq > 26.7 is about 0.2 Sv. This agrees withinthe error with the results of Gladyshev et al. [2000], whoestimated the removal rates for the NWS between July andAugust 1997 and for the NS between May and August1997. For these removal rates, we estimate that about 2 �103 km3 of the DSW with a density of 26.7 sq does notleave the northern shelves until December. Most of this lateDSW lying at depths shallower than about 150 m (the lowerlimit of winter convection according to Moroshkin [1966])is subject to winter overturning on the shelf, eventuallycompletely mixing with the surrounding water and becom-ing part of the next year’s new DSW. Because of winterconvection the northern DSW flux between November andFebruary would be close to zero.

5.2. Seasonal Variability of the Okhotsk Sea ModeWater

[50] Historical data for April–May (spring) and Septem-ber (fall) as representative months are used to show seasonalvariability of the OSMW (Figure 14). April and May arecombined because of the few observations in early springand because we require at least 7 observations within each1� square. We display the averaged anomalies of tempera-ture at 26.9 sq, to highlight the impact of the northern DSW,which is the only source of freezing water that ventilates thedeep sea at this density. In general, the distribution andpropagation of the anomalies in the OSMWare consistent atall densities (not shown).

[51] In April–May (Figure 14a), a large cold anomalywith (q < �0.4�C) exists in the northwestern Kuril Basin,apparently reflecting how rapidly the ESC transports theMDSW southward. A separate cold anomaly is also appar-ent in the vicinity of Bussol’ Strait, where negative anoma-lies persist into June–July (Figures 14b and 14c). Thestrong cold anomaly (q < �0.4�C) remains near the Sakha-lin coast because of DSWoutflow from the northern shelvesin June–July (not shown). Beginning in August, warmanomalies rapidly develop in the OSMW (Figure 14d). Alarge positive anomaly (q > 0.2�C) grows along the entireSakhalin slope indicating that transport of the MDSW to thedeep sea dramatically reduces. This anomaly persists intoSeptember–November (Figures 14e and 14f). For the sameperiod, the large positive anomaly quickly grows nearBussol’ Strait, occupying about half of the Kuril Basin inSeptember.[52] In summary, there are two well-distinguished seasons

in the OSMW. The cold season occurs between April andJuly, when the MDSW produces large cold anomalies in thewestern Okhotsk Sea. Cold anomalies also occur in theeastern Okhotsk between April and July, indicating rela-tively weak inflow of the MNPW to the Okhotsk Sea. Thewarm season occurs between August and November, whenthe DSW flux from the shelves strongly decreases and waterexchange between the Okhotsk Sea and North Pacificintensifies.[53] Seasonal variability of the thermohaline structure

through the entire OSMW in the ESC and Bussol’ Strait isshown in Figure 15, using data averaged within areas ESC

Figure 13. The average distribution of potential density at the bottom on the northern Okhotsk Seashelves in (a) May–June and (b) September–October based on the historical data. The shaded areasdenote regions where the bottom density exceeds 26.7 sq. The thin line is the 200 m isobath.

Table 4. May–June (VSP) and September–October (VF) Volumes

of the Dense Shelf Water (DSW), Their Average Thickness (H),

and Summer–Fall Removal Rate (U) on the Okhotsk Sea Northern

Shelves

sq VSP, km3 � 103 VF, km

3 � 103 H, m U, Sv

>26.7 5.2 ± 0.3 3.2 ± 0.2 57 ± 29 0.19 ± 0.08>26.8 1.4 ± 0.1 0.4 ± 0.1 41 ± 22 0.1 ± 0.04>26.9 0.24 ± 0.02 . . . 29 ± 16 0.02 ± 0.01


and BS, encompassed by dashed lines in Figure 1. April–May and August–September were selected as representa-tive months. Within the OSMW (sq = 26.7–27.0) thecoldest temperature occurs in April–May because ofintensive DSW injection from the shelves (Figure 15).The temperature then gradually increases until September,when the largest amount of the MNPW enters the OkhotskSea (Figure 15b). It appears that the variations of temper-ature below the OSMW layer (sq = 27.1–27.4) are alsorelated to the North Pacific inflow. This result is consistent

with observations of Wakatsuchi and Martin [1991], al-though we propose a different reason for the variability ofthis denser water, which is sometimes referred to as lowerOkhotsk Intermediate Water. We average gridded values oftemperature and salinity south of 49�N (37 gridded points)to estimate the seasonal variability of OSMW in the KurilBasin. The average seasonal fluctuations in the OSMW arelargest at 26.7 sq, with a range of about 0.7�C, andapparently penetrating to sq = 27.0 (about 500 m depth),with a range of about 0.2�C (Figure 15a). The former is

Figure 14. Averaged potential temperature anomalies relative to the annual mean in the deep OkhotskSea at 26.9 sq in (a) April–May, (b) June, (c) July,(d) August, (e) September, and (f) October–November.Shading indicates the statistically significant (95%) anomalies. Solid contours are positive anomalies, anddashed contours are negative anomalies. Contour intervals are 0.2�C.


significantly different from zero at the 95% confidencelevel. In contrast to temperature, the average seasonalfluctuations in salinity on isopycnals are quite small southof 49�N with small standard deviation (Figure 15). Withinthe OSMW (26.7 sq), the maximum annual salinity rangeis 0.05 psu, but at 27.0 sq, the range is only 0.01 psu orless than the accuracy of the salinity measurements.

5.3. Seasonal Cycle of Okhotsk Sea Mode Waterin the Kuril Basin

[54] The average temperature of the OSMW in the KurilBasin increases from May to September and then is un-changed through December (Figure 16). (Only the Prof.Levanidov 1995 cruise is used for December.) The lattermeans that a relatively weak DSW flux of about 0.1–0.2 Svfrom the northern shelves (Table 4) might be equilibrated bythe NPW inflow in the fall. This situation might continueuntil February, when the convective DCW would begin todecrease the temperature of the OSMW. The temperature

would continue to drop until May because of growing DSWflux from the Sakhalin and Terpeniya Bay polynyas andfurther from the northern shelves. This hypothesis issketched in Figure 16.[55] To describe the OSMW seasonal cycle in the Kuril

Basin we use a simple box model for the 26.7–27.0 sqlayer. We limit our calculation to south of 49�N because ofthe scarcity of data to the north due to ice cover in spring.However, this limitation is not important because mostDSW descending from the shelves eventually flows to theKuril Basin south of this latitude. Temperature in theOSMW is governed by

V@T

@t¼ q1T1 þ q2T2 � q1 þ q2ð ÞT3; ð2Þ

where V is the OSMW volume south of 49�N, q1 is theDCW/DSW volume transport, including convective DCWin the Kuril Basin (only for Case III below) and the DSWfrom the Sakhalin and northern shelves, q2 is the NPWvolume transport, T1 is the freezing temperature of theDCW in the Kuril Basin and the DSW from the shelves, T2is the average temperature of the NPW for 26.7–27.0 sq, T3is the average temperature of the OSMW, @T@t is the OSMWtemperature fluctuation, and q1 + q2 is the OSMWproduction rate. We assume linear increase (decrease) oftemperature in the OSMW in summer (in late winter andspring). We do not use salinity because its summer increasein the OSMW is only 0.03 psu or just slightly exceeding thesalinity accuracy. From the data, T1 = �1.7 ± 0.06�C, T2 =2.93 ± 0.55�C, and T3 = 0.99 ± 0.17�C. The volume of theOSMW south of 49�N is 9.5 ± 0.5 � 104 km3, based on thedepths of 26.7–27.0 sq isopycnal surfaces and ETOP05topography.[56] Case I: Summer. From June to September (about 120

days) the average temperature of OSMW increases by0.37�C from 0.75 ± 0.24�C to 1.12 ± 0.21�C (Figure 16).Since most of Sakhalin DSW leaves the shelf before May–June (Figure 7), q1 is defined to a large degree by thesummer flux of DSW from the northern shelves (Table 4).Hence the only unknown parameter in (2) is q2. Solving (2)

Figure 15. Average potential temperature/salinity curvesfor (a) ESC and (b) BS. Both regions are encompassed bydashed lines in Figure 1. April–May (solid) and August–September (dashed) curves show the seasonal variations inthe region of the East Sakhalin Current and near Bussol’Strait based on the historical data. Vertical and horizontalbars represent the standard deviation about the means oftemperature and salinity, respectively.

Figure 16. Monthly average potential temperature ofOSMW in the Kuril Basin (26.7–27.0 sq, south of 49�N)based on the gridded historical data. Vertical bars are thestandard deviation of the mean. The solid curve is a linearfit to temperature between May and December. The dashedcurve is hypothesized behavior of temperature betweenDecember and May.


yields q2 = 2.0 ± 1.6 Sv. The known errors in this estimateinclude temperature uncertainty estimated from standarddeviations, a large error from @T

@t , the q1 uncertainty givenin Table 4, and the error associated with the OSMW volumecalculation. Our estimate of summer OSMW productionrate is 2.2 ± 1.7 Sv.[57] Case II: Winter. Temperature does not change be-

tween September and December, so V@T@t = 0. The solution of

(2) thus is q1q2= 0.7. Assuming that q1 drops from 0.2 Sv in

October to almost zero in November–December, because itmixes by winter convection, and there is no new DSWformation on the shelves, we take q1 = 0.1 Sv. This yields q2 0.1 ± 0.1 Sv. Thus fall–winter OSMW production (q1 +q2) is estimated as 0.2 ± 0.1 Sv or much less than insummer.[58] Case III: Spring. The average temperature of OSMW

decreases between February and May (Figure 16), compen-sating its summer increase. To account for the ForerunnerSW contribution between March and May [Takizawa, 1982;Watanabe and Wakatsuchi, 1998] to seasonal transforma-tion of the OSMW, we rewrite (2) as

V@T

@t¼ q1T1 þ q2T2 þ q3T4 � q1 þ q2 þ q3ð ÞT3; ð3Þ

where q3 is the Forerunner SW volume transport and T4 isthe average temperature of Forerunner SW in 26.7–27.0 sq.We take T4 = 3.94 ± 0.53�C from our data and q3 = 0.2 Svfollowing Watanabe and Wakatsuchi [1998]. Now there arealso two unknown parameters q1 and q2, which cannot bederived from our data. Our principal interest is to estimatethe spring DSW production rate, which would cause thehypothesized OSMW cooling. Assuming that the springNPW flux q2 is the same as winter (Case II), the solution forq1 is 1.5 ± 1.2 Sv. Separating the DSW from the Kuril BasinDCW, using oxygen and silicate properties, gives for theformer 1.4 ± 1.1 Sv, which is greater than the 0.8 ± 0.3 Svestimate of Gladyshev et al. [2000] for spring–summer1997, for the NWS only. Given q1, q2, and q3, the springOSMW production rate is 1.8 ± 1.3 Sv. We also solve theequation for different NPW flux values (Table 5), where 0.1Sv is subtracted from q1 to exclude the Kuril Basin DCWcontribution. Increasing the spring NPW flux up to 0.5 (1.0)Sv requires a corresponding increase in the DSW produc-tion up to 1.7 (2.1) Sv. These values might overestimate thespring DSW production rate on the Okhotsk Sea shelves.[59] Taking 1.4 Sv as a suitable value for the spring DSW

production, an annual average DSW flux and OSMWproduction are estimated as well as a residence time ofthe OSMW in the Kuril Basin. Averaging the results for the

Cases I–III, the annual average DSW flux from the shelvesis 0.6 ± 0.6 Sv, while the annual average OSMW productionin the Kuril Basin is 1.4 ± 1.2 Sv. The latter is comparable tothe 1.0 ± 0.2 Sv of Gladyshev et al. [2000] for 26.7–26.9sq, and to the 2.0 ± 1.5 Sv of Yasuda [1997] for lowpotential vorticity Oyashio water. However, it is less thanthe 3.3 Sv of Wong et al. [1998].[60] The residence time of the OSMW in the Kuril Basin

is defined as

t ¼ V

FOSMW

; ð4Þ

where t is the residence time, V is the OSMW volume as in(2), and FOSMW is the annual average OSMW productionrate in the Kuril Basin. This yields a residence time of 2 ±1.7 years. This is consistent with the 1.4 year estimate ofWong et al. [1998] based on CFC data for sq < 27.0. Itdiffers considerably from the 4–14 years of Gladyshev et al.[2000]. The latter did not include the contribution ofSakhalin DSW or of Kuril Basin DCW, although theyoverestimated the contribution of Shelikhov DSW. Inaddition, DSW outflow from the NWS starts at least at thebeginning of March, not in April as was expected byGladyshev et al. [2000]. Rapid winter–spring outflow fromthe shelves may be due to entrainment of newly formedDSW in polynyas, which forces the DSW formed earlier toflow faster. Shapiro and Hill [1997] have shown thatentrainment is able to accelerate the DSW spreading on theshelves by a factor of 2.

6. Conclusion

[61] Using Russian historical data and recently completedCTD surveys from the Okhotsk Sea, the formation andsubsequent expansion of dense shelf water that forms incoastal polynyas during winter has been described and thepotential for open water deep convection has been evaluat-ed. In spring, most of the dense shelf water drains rapidlyfrom the northwestern and Sakhalin shelves to the deepersea and is transported by the East Sakhalin Current to theKuril Basin. It was shown that Shelikhov Bay is not animportant site of OSMW ventilation, likely because ofvigorous tidal mixing. It is also shown that the DSWdescending from the northern shelves occurs most of theyear. Winter transformation of the Soya Water can produce aconvected water with density greater than 26.7 sq in thesoutheastern Kuril Basin, providing another, minor contri-bution to OSMW transformation. This cold input as well aswater exchange with the North Pacific leads to largeseasonal fluctuations of temperature in the OSMW (sq =26.7–27.0), while the seasonal salinity changes are signif-icantly smaller.[62] Using the observed seasonal variations in tempera-

ture in the Kuril Basin and applying a simple isopycnal boxmodel, the summer OSMW production rate is estimated tobe 2.2 ± 1.7 Sv. The OSMW production rate decreasesconsiderably to about 0.2 ± 0.1 Sv in winter. Assuming thattemperature of the OSMW decreases from February throughMay to compensate its summer increase, we estimate that atleast 1.4 ± 1.2 Sv of total DSW volume transport isnecessary to provide these changes. The summer DSW

Table 5. Estimated February–May DSW Volume Transport

(qDSW) for Given North Pacific Inflow (q2) in the OSMW Density

Range

q2 qDSW, Sv

0.1 1.4 ± 1.30.3 1.6 ± 1.40.5 1.7 ± 1.50.7 1.9 ± 1.70.9 2.0 ± 1.81.0 2.1 ± 1.9


volume transport is about 0.2 ± 0.1 Sv. The residence timeof the OSMW in the Kuril Basin is 2 ± 1.7 years. Thisestimate is consistent with Wong et al. [1998] but signifi-cantly shorter then that of Gladyshev et al. [2000] for 26.8–26.9 sq.[63] Winter observations on the Sakhalin shelf and early

spring measurements along the southern Kuril Islandswould be very useful for better estimating the SakhalinDSW production and to better quantify the contribution ofwinter convection. Monitoring properties and currents in theneck of Shelikhov Bay is also desirable. Because tidalmixing is crucial for Okhotsk Mode Water formation, itshould be included in theoretical models.

[64] Acknowledgments. The National Climatic Data Center providedthe precipitation data. We also thank Lena Dmitrieva for providing part ofthe Russian data set. We thank Steve Riser and Kensuke Takeuchi forleading the 1995 and 1998 cruises. We are also grateful to TakatoshiTakizawa of JAMSTEC for his support on 1998 and 1999 Prof. Khromovruises. Comments by two anonymous reviewers and assistance from A.Shcherbina were valuable. This work was supported by funds from CoreResearch for Evolutional Science and Technology, Japanese Science andTechnology Corporation (SVG and MW), and NSF OCE-9811958 (LDT).

ReferencesAlfultis, M. A., and S. Martin, Satellite passive microwave studies of theSea of Okhotsk ice cover and its relation to oceanic processes, 1978–1982, J. Geophys. Res., 92, 13,013–13,028, 1987.

Aota, M., Studies on the Soya Warm Current (in Japanese), Low Temp. Sci.,Ser. A, 33, 151–172, 1975.

Bobkov, A. A., New schema of the circulation and water mass distributionin the southern Kuril area, Umi Sora, 69, 41–52, 1993.

Boyer, T. P., and S. Levitus, Quality control and processing of historicaltemperature, salinity and oxygen data, 65 pp., NOAA Tech. Rep. NESDIS81, Natl., Oceanic and Atmos. Admin., Silver Spring, Md., 1994.

Chapman, D. C., and G. Gawarkiewicz, Shallow convection and buoyancyequilibration in an idealized coastal polynya, J. Phys. Oceanogr., 27,555–566, 1997.

Favorite, F., A. J. Dodimead, and K. Nasu, Oceanography of the SubarcticPacific Region, 1960–1971, vol. 33, 187 pp., Int. N. Pac. Fish. Comm.,Vancouver, B. C., Can., 1976.

Freeland, H. J., A. S. Bychkov, F. Whitney, C. Taylor, C. S. Wong, and G. I.Yurasov, WOCE section P1W in the Sea of Okhotsk: 1. Oceanographicdata description, J. Geophys. Res., 103, 15,613–15,623, 1998.

Gladyshev, S. V., S. Martin, S. C. Riser, and A. L. Figurkin, Dense waterproduction on the northern Okhotsk shelves: Comparison of ship-basedspring–summer observations for 1996 and 1997 with satellite observa-tions, J. Geophys. Res., 105, 26,281–26,299, 2000.

Itoh, M., Formation and distribution of the Okhotsk Sea Intermediate Water,Ph. D. thesis, 59 pp., Hokkaido Univ., Hokkaido, Jpn., 2000.

Kimura, N., and M. Wakatsuchi, Processes controlling the advance andretreat of sea ice in the Sea of Okhotsk, J. Geophys. Res., 104,11,137–11,150, 1999.

Kitani, K., An oceanographic study of the Okhotsk Sea-particularly inregard to cold waters, Bull. Far Seas Fish. Res. Lab., 9, 45–77, 1973.

Kowalik, Z., and I. Polyakov, Tides in the Sea of Okhotsk, J. Phys. Ocea-nogr., 28, 1389–1409, 1998.

Martin, S., R. Drucker, and K. Yamashita, The production of ice and denseshelf water in the Okhotsk Sea polynyas, J. Geophys. Res., 103, 27,771–27,782, 1998.

Moroshkin, K. V., Water masses of the Sea of Okhotsk, Joint Publ. Res.Serv. 43942, 98 pp., U.S. Dep. of Commer., Washington, D. C., 1966.

Nakamura, T., T. Awaji, T. Hatayama, K. Akitomo, and T. Takizawa, Tidalexchange through the Kuril Straits, J. Phys. Oceanogr., 30, 1622–1644,2000a.

Nakamura, T., T. Awaji, T. Hatayama, K. Akitomo, T. Takizawa, T. Kono,Y. Kawasaki, and M. Fukasawa, The generation of large-amplitude un-steady lee waves by subinertial K1 tidal flow: A possible vertical mixingmechanism in the Kuril Straits, J. Phys. Oceanogr., 30, 1601–1621,2000b.

Riser, S. C., Exchange of water between the Okhotsk Sea and North PacificOcean through the Kurile Straits, paper presented at International Work-shop on the Okhotsk Sea and Arctic, Jpn. Mar. Sci. and Technol., Tokyo,1996.

Riser, S. C., G. I. Yurasov, and M. J. Warner, Hydrographic and tracermeasurements of the water mass structure and transport in the OkhotskSea in early spring, in Proceedings of the Workshop on the Okhotsk Seaand Adjacent Areas, PICES Sci. Rep. 6, edited by Y. Nagata, V. B.Lobanov, and L. D. Talley, pp. 138–143, North Pac., Mar. Sci. Org.,Sidney, B. C., Can., 1996.

Rogachev, K. A., Recent variability in the Pacific western subarctic bound-ary currents and Sea of Okhotsk, Prog. Oceanogr., 47, 299–336, 2000.

Sagalevich, A. M., and V. N. Ivanenkov, High-turbidity layers in the TinroBasin, Sea of Okhotsk, Trans. Akad. Sci., 267, 215–217, 1982.

Shapiro, G. I., and A. E. Hill, Dynamics of dense water cascades at the shelfedge, J. Phys. Oceanogr., 27, 2381–2394, 1997.

Takizawa, T., Characteristics of the Soya Warm Current in the Okhotsk Sea,J. Oceanogr. Soc. Jpn., 38, 281–292, 1982.

Talley, L. D., An Okhotsk water anomaly: Implications for ventilation in theNorth Pacific, Deep Sea Res., Part A, 38, Suppl. 1, 171–190, 1991.

Talley, L. D., Distribution and formation of the North Pacific IntermediateWater, J. Phys. Oceanogr., 23, 517–537, 1993.

Talley, L. D., , and Y. Nagata (Eds.), The Okhotsk Sea and Oyashio Region,PICES Sci. Rep. 2, 227 pp., North Pac. Mar. Sci. Org., Sidney, B. C.,Can., 1995.

Wakatsuchi, M., and S. Martin, Water circulation in the Kuril Basin of theOkhotsk Sea and its relation to eddy formation, J. Oceanogr. Soc. Jpn.,47, 152–168, 1991.

Watanabe, K., On the reinforcement of the East Sakhalin Current precedingto the sea ice season off the coast of Hokkaido, Oceanogr. Mag., 14,117–130, 1963.

Watanabe, T., and M. Wakatsuchi, Formation of 26.8–26.9 sq water in theKuril Basin of the Sea of Okhotsk as a possible origin of North PacificIntermediate Water, J. Geophys. Res., 103, 2849–2865, 1998.

Wong, C. S., R. G. Matear, H. J. Freeland, F. A. Whitney, and A. S.Bychkov, WOCE line P1W in the Sea of Okhotsk: 2. CFCs and theformation rate of intermediate water, J. Geophys. Res., 103, 15,625–15,642, 1998.

Yasuda, I., The origin of the North Pacific Intermediate Water, J. Geophys.Res., 102, 893–909, 1997.

You, Y., N. Suginohara, M. Fukasawa, I. Yasuda, I. Kaneko, H. Yoritaka,and M. Kawamiya, Roles of the Okhotsk Sea and Gulf of Alaska informing the North Pacific Intermediate Water, J. Geophys. Res., 105,3253–3280, 2000.

��S. Gladyshev and M. Wakatsuchi, Institute of Low Temperature Science,

Hokkaido University, Sapporo 060-0819, Japan.G. Kantakov, Sakhalin Research Institute of Fisheries and Oceanography,

Yuzhno-Sakhalinsk 693016, Russia.G. Khen, Pacific Scientific Research Fisheries Center, Vladivostok

690600, Russia.L. Talley, Scripps Institution of Oceanography, University of California,

San Diego, La Jolla, CA 92093–0230, USA.


distribution, formation, and seasonal variability of...

Documents